Model comparison
Claude 3.5 Haiku vs DeepSeek V4.1 Flash
DeepSeek V4.1 Flash is the stronger model overall, scoring 52.8 to 29.2 on the Noometry Index.
Last verified . 25 shared benchmarks.
DeepSeekSummary
- They share 25 benchmarks with published results for both. Claude 3.5 Haiku scores higher in 0 categories and DeepSeek V4.1 Flash in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4.1 Flash leads 66.7 to 14.7.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 4.3% for Claude 3.5 Haiku and 98.3% for DeepSeek V4.1 Flash.
- DeepSeek V4.1 Flash has downloadable open weights; the other is API-only.
Side by side
| Claude 3.5 Haiku | DeepSeek V4.1 Flash | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 29.2 | 52.8 |
| Released | 2024-10-22 | 2026-09-09 |
| Weights | Proprietary | Open |
| Context window | — | 1M |
| Max output | — | 393K |
| Input $ / M tokens | — | $0.15 |
| Output $ / M tokens | — | $0.60 |
| Results tracked | 49 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 32.9 (#265), DeepSeek V4.1 Flash: 52.9 (#32)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| SciCode | 27.4% | 51.9% |
| LMArena Coding | 1286 | 1506 |
| Aider Polyglot | 28% | — |
| LMArena WebDev | — | 1619 |
| WeirdML | 30.7% | — |
| BigCodeBench Instruct | 46.1% | — |
| LiveBench Coding | 51.4% | — |
| BigCodeBench Complete | 59% | — |
| CadEval | 32% | — |
| ALE-Bench | — | 1,092 |
Agentic & Tool Use DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 28.0 (#95), DeepSeek V4.1 Flash: 31.2 (#69)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| APEX-Agents | — | 39.5% |
| BALROG | 19.3% | — |
| GDP.pdf | — | 19.8% |
Reasoning DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 17.7 (#290), DeepSeek V4.1 Flash: 50.2 (#36)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| CritPt | 0% | 14.3% |
| LMArena Hard Prompts | 1251 | 1483 |
| DTBench | 56.7% | 89.9% |
| Epoch Capabilities Index | 127.15 | 154.9 |
| NYT Connections (extended) | — | 89.6% |
| LiveBench Reasoning | 28.1% | — |
| Mystery Game Puzzles | — | 43% |
| LiveBench Data Analysis | 48.5% | — |
| LMCA | — | 47% |
| Surface Evolver Bench | — | 46.3% |
| LiveBench | 43.5% | — |
Math DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 14.7 (#300), DeepSeek V4.1 Flash: 66.7 (#25)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 4.3% | 98.3% |
| LMArena Math | 1244 | 1477 |
| FrontierMath (Tiers 1-3) | — | 67.4% |
| FrontierMath Tier 4 | — | 26.8% |
| ProofBench | — | 54% |
| Omni-MATH | 22.4% | — |
| LiveBench Math | 35.5% | — |
| MATH Level 5 | 46.4% | — |
| FrontierMath (Feb 2025 set) | 0.3% | — |
Knowledge DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 18.7 (#281), DeepSeek V4.1 Flash: 57.9 (#38)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| GPQA Diamond | 38.1% | 89.8% |
| LMArena Expert | 1208 | 1506 |
| MMLU-Pro | 60.5% | — |
| Confabulations | 36.7% | — |
| GPQA (HELM) | 36.3% | — |
| MMLU | 74.3% | — |
Multimodal DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 26.8 (#117), DeepSeek V4.1 Flash: 39.1 (#61)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| LMArena Vision | 1092 | 1277 |
| GeoBench | 34% | — |
| Furniture Assembly | — | 34.2% |
Multilingual DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 40.0 (#218), DeepSeek V4.1 Flash: 55.0 (#35)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| LMArena Non-English | 1238 | 1448 |
| LMArena Chinese | 1229 | 1497 |
| LMArena French | 1264 | 1452 |
| LMArena German | 1237 | 1484 |
| LMArena Japanese | 1175 | 1412 |
| LMArena Korean | 1173 | 1452 |
| LMArena Russian | 1253 | 1471 |
| LMArena Spanish | 1261 | 1459 |
Instruction Following DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 62.9 (#234), DeepSeek V4.1 Flash: 77.3 (#26)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| LMArena Instruction Following | 1241 | 1474 |
| LiveBench Instruction Following | 61.9% | — |
| IFEval | 79.2% | — |
Long Context DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 38.3 (#200), DeepSeek V4.1 Flash: 45.2 (#47)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| LMArena Longer Query | 1261 | 1475 |
Writing & Preference DeepSeek V4.1 Flash leads
Claude 3.5 Haiku: 42.7 (#234), DeepSeek V4.1 Flash: 65.4 (#48)
| Benchmark | Claude 3.5 Haiku | DeepSeek V4.1 Flash |
|---|---|---|
| LMArena Text | 1255 | 1462 |
| LMArena Creative Writing | 1233 | 1435 |
| EQ-Bench Creative Writing | 1146 | 1540 |
| LMArena Multi-Turn | 1265 | 1457 |
| Short-Story Creative Writing | 73.5% | — |
| WildBench | 76% | — |
| LiveBench Language | 35.4% | — |
Frequently asked questions
Is Claude 3.5 Haiku better than DeepSeek V4.1 Flash?
DeepSeek V4.1 Flash is the stronger model overall, scoring 52.8 to 29.2 on the Noometry Index.
Is Claude 3.5 Haiku or DeepSeek V4.1 Flash better for coding?
DeepSeek V4.1 Flash scores higher on coding benchmarks: 52.9 versus 32.9 in the Noometry coding category.
How many benchmarks do Claude 3.5 Haiku and DeepSeek V4.1 Flash share?
25 benchmarks have published results for both models. Claude 3.5 Haiku has 49 scored results on Noometry and DeepSeek V4.1 Flash has 37.